Be the first to post a text review of PyCV. Rate and review a project by clicking thumbs up or thumbs down in the right column.

View all reviews

• Code committed

  mtpham committed revision 126 to the PyCV SVN repository, changing 12 files

  posted by mtpham 52 days ago

• Development Release: PyCV 0.3.0

  .. -*- rest -*- ========================================================================================= PyCV - A Computer Vision Package for Python Incorporating Fast Training of Face Detection ========================================================================================= :Website: http://www.ntu.edu.sg/home5/pham0004/pycv/ :Authors: Minh-Tri Pham <mtpham@ntu.edu.sg>, Viet-Dung D. Hoang <hken@pmail.ntu.edu.sg>, Tat-Jen Cham <astjcham@ntu.edu.sg> :Version: 0.3.0 (devel) :Released: September 2008 .. Contents:: What's New? =========== Release notes for version 0.3.0 -------------------------------- This release includes two new classification methods: decision stump and our Multi-exit Asymmetric Boosting (CVPR'08). I'm in the process of converting every train_xxxx() function into Machine Learner classes. This way, the whole package is more robust and easily extendible. The pycv.cs.ml package has been received most of our attention. In the next release, we will improve the pycv.cs.cv package significantly. Changes from version 0.2.2 to version 0.3.0 ------------------------------------------- - Added class BCObjective to model an objective function of a binary classifier. - Upgraded PyCV to be compatible with numpy 1.2 testing modules. Now it requires package 'nose' installed to test (SciPy's decision). See: http://www.scipy.org/scipy/scipy/wiki/TestingGuidelines for more details. - Added class MachineLearner to represent a machine learner. - Added class BasicRegStats and class Stats2XY for regression. - Added class LinearRegressorLearner, rendering method train_LinearRegessor() obsolete. - Added class NBClassifierLearner, rendering method train_NBClassifier() obsolete. - Added a test on learning a NBClassifier. - Moved PredictPdfInterface from pycv.cs.ml to pycv.cs.ml.cla. - Added class LDAClassifierLearner. - Removed method train_LDA(). obsolete. - Added some tests for learning an LDA classifier. - Improved method find_classification_threshold() more robust. - Added class ShifterLearner, renderingmethod train_Shifter() obsolete. - Found a bug with SSE2 and method cblas_dsum(), currently disabled sse2 in method plat_det(). - Added tests for method thresh_normal_1d() - Added class L1DAClassifier and an abstract of its learner, to replace UAC (our unpublished classifier). - Added class L1DClassifier and class L1DClassifierLearner. - Created a sandox folder to store experimenting work. - Moved class UAC and other related subroutines to my sandbox. - Added class AdaBoost to replace method train_AdaBoost(). - Added class AsymBoostVJ to replace method train_VJ(). - Added class OnlineBoosting to partially replace the current train_DBC() method. - Added a test for class OnlineBoosting. - Added class AsymBoostPC to replace method train_PC(). - Added tests for class AsymBoostPC. - Added class OnlineAsymBoostPC and a test for it. - Removed class DirectFilteringCascade and method train_DFC(). - Added class CascadeLearner to represent how a traditional cascade is learned. - Added class MultiexitAsymBoost to represent our Multi-exit Asymmetric Boosting learning algorithm. - Added a test for MultiexitAsymBoost. - Added classes Stump, StumpLearner, and their tests. - Added class GlobalParams to group all global parameters of pycv. - Fixed some bugs. - Added class Visualizer to deal with visualizing data. - Made method stats_integral() in haar.py as a private method - Created sub-package named 'features' in pycv.cs.cv and moved haar.py to the package. - Added VideoCapturer to capture images. By default, pycv uses OpenCV's highgui to capture images from a device. However, if package VideoCapture (Python) is available, it'll use the package instead for fast and large image capture (using DirectShow). - Added PCA and remove_direction in linalg package - Created package pycv.cs.visual and moved every class related to Visualizer to this package. - Added CvFont to CVTypes, now cvInitFont() and cvPutText() is working. - Added a large number of CvXYZ.. class wrappers to CVtypes.py.

  posted by mtpham 434 days ago

• Development Release 0.3.0 file released: pycv-0.3.0.zip

  - Added class BCObjective to model an objective function of a binary classifier. - Upgraded PyCV to be compatible with numpy 1.2 testing modules. Now it requires package 'nose' installed to test (SciPy's decision). See: http://www.scipy.org/scipy/scipy/wiki/TestingGuidelines for more details. - Added class MachineLearner to represent a machine learner. - Added class BasicRegStats and class Stats2XY for regression. - Added class LinearRegressorLearner, rendering method train_LinearRegessor() obsolete. - Added class NBClassifierLearner, rendering method train_NBClassifier() obsolete. - Added a test on learning a NBClassifier. - Moved PredictPdfInterface from pycv.cs.ml to pycv.cs.ml.cla. - Added class LDAClassifierLearner. - Removed method train_LDA(). obsolete. - Added some tests for learning an LDA classifier. - Improved method find_classification_threshold() more robust. - Added class ShifterLearner, renderingmethod train_Shifter() obsolete. - Found a bug with SSE2 and method cblas_dsum(), currently disabled sse2 in method plat_det(). - Added tests for method thresh_normal_1d() - Added class L1DAClassifier and an abstract of its learner, to replace UAC (our unpublished classifier). - Added class L1DClassifier and class L1DClassifierLearner. - Created a sandox folder to store experimenting work. - Moved class UAC and other related subroutines to my sandbox. - Added class AdaBoost to replace method train_AdaBoost(). - Added class AsymBoostVJ to replace method train_VJ(). - Added class OnlineBoosting to partially replace the current train_DBC() method. - Added a test for class OnlineBoosting. - Added class AsymBoostPC to replace method train_PC(). - Added tests for class AsymBoostPC. - Added class OnlineAsymBoostPC and a test for it. - Removed class DirectFilteringCascade and method train_DFC(). - Added class CascadeLearner to represent how a traditional cascade is learned. - Added class MultiexitAsymBoost to represent our Multi-exit Asymmetric Boosting learning algorithm. - Added a test for MultiexitAsymBoost. - Added classes Stump, StumpLearner, and their tests. - Added class GlobalParams to group all global parameters of pycv. - Fixed some bugs. - Added class Visualizer to deal with visualizing data. - Made method stats_integral() in haar.py as a private method - Created sub-package named 'features' in pycv.cs.cv and moved haar.py to the package. - Added VideoCapturer to capture images. By default, pycv uses OpenCV's highgui to capture images from a device. However, if package VideoCapture (Python) is available, it'll use the package instead for fast and large image capture (using DirectShow). - Added PCA and remove_direction in linalg package - Created package pycv.cs.visual and moved every class related to Visualizer to this package. - Added CvFont to CVTypes, now cvInitFont() and cvPutText() is working. - Added a large number of CvXYZ.. class wrappers to CVtypes.py.

  posted 434 days ago

• File released: /Development Release/0.3.0/pycv-0.3.0.zip

  posted 434 days ago

• Stable Release 0.2.2 file released: pycv-0.2.2.win32-py2.5.exe

  the same as 0.2.1 with some changes in copyright notices

  posted 632 days ago

• Stable Release 0.2.2 file released: pycv-0.2.2.zip

  the same as 0.2.1 with some changes in copyright notices

  posted 632 days ago

• Stable Release 0.2.2 file released: pycv_demo-0.2.2.zip

  the same as 0.2.1 with some changes in copyright notices

  posted 632 days ago

• Stable Release: PyCV 0.2.2

  the same as 0.2.1 with some changes in copyright notices

  posted by mtpham 632 days ago

• File released: /Stable Release/0.2.2/pycv-0.2.2.win32-py2.5.exe

  posted 632 days ago

• File released: /Stable Release/0.2.2/pycv-0.2.2.zip

  posted 632 days ago